From c226ded799ff2b9a181d38ac0d88ac88365c35b1 Mon Sep 17 00:00:00 2001
From: luccioman <luccioman@users.noreply.github.com>
Date: Tue, 30 May 2017 12:32:14 +0200
Subject: [PATCH] Fix unescape of URLs having some '%' chars but not
 percent-encoded

---
 .../net/yacy/cora/document/id/MultiProtocolURL.java   | 11 ++++++++++-
 .../yacy/cora/document/id/MultiProtocolURLTest.java   |  1 +
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/source/net/yacy/cora/document/id/MultiProtocolURL.java b/source/net/yacy/cora/document/id/MultiProtocolURL.java
index 7a42826a2..2b155b6ce 100644
--- a/source/net/yacy/cora/document/id/MultiProtocolURL.java
+++ b/source/net/yacy/cora/document/id/MultiProtocolURL.java
@@ -688,7 +688,7 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
      * Decodes a <code>application/x-www-form-urlencoded</code> string using UTF-8 encoding.
      *
      * @param s the string to decode
-     * @return the newly decoded string
+     * @return the newly decoded string, or the original string when it doesn't match the <code>application/x-www-form-urlencoded</code> format
      */
     public static String unescape(final String s) {
     	try {
@@ -697,6 +697,15 @@ public class MultiProtocolURL implements Serializable, Comparable<MultiProtocolU
 			/* This should not happen */
 			ConcurrentLog.logException(e);
 			return s;
+		} catch(Exception e) {
+			/*
+			 * URLDecode may throw an IllegalArgumentException (or any other
+			 * Exception in future implementations) when the string doesn't
+			 * match the application/x-www-form-urlencoded format: in that case
+			 * return the original string. 
+			 * Example case : when the valid '%' character is used in a URL but without percent encoding purpose.
+			 */
+			return s;
 		}
     }
 
diff --git a/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java b/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
index a2ba443f5..482ccc2c7 100644
--- a/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
+++ b/test/java/net/yacy/cora/document/id/MultiProtocolURLTest.java
@@ -368,6 +368,7 @@ public class MultiProtocolURLTest {
 				new String[] { "unencoded rfc3986 unreserved ascii chars:-.~_",
 						"unencoded rfc3986 unreserved ascii chars:-.~_" },
 				new String[] { "http://simpleurl.com/", "http://simpleurl.com/" },
+				new String[] { "http://not-a-x-www-form-urlencoded.com/example.php?params=a|b&paramwithpercent=%param%", "http://not-a-x-www-form-urlencoded.com/example.php?params=a|b&paramwithpercent=%param%" },
 				new String[] {
 						"http://url-with-unencoded-query-and-anchor.net/path?q=asciiquery&p1=param1&p2=pâräm2&p3=简化字#anchor",
 						"http://url-with-unencoded-query-and-anchor.net/path?q=asciiquery&p1=param1&p2=pâräm2&p3=简化字#anchor" }, };